Domain-Specific Sentiment Classifier Using Segmented Lexicons
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sentiment classification methods fail to accurately account for domain-specific words and concepts, leading to inconsistent sentiment analysis across different domains, as words can have positive or negative connotations depending on the context.
Innovation Solution
A method and system for constructing a domain-specific sentiment classifier by scoring and identifying high-sentiment documents using a domain-independent sentiment lexicon, creating a domain-specific lexicon, and generating a classifier to accurately classify sentiment in specific domains.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a domain-independent sentiment lexicon is used for sentiment classification, then the classification can be applied across multiple domains, but the accuracy deteriorates because words have different sentiments in different domains
Solution Approach 1:
The patent segments the sentiment classification system into domain-independent and domain-specific components. The domain-independent lexicon provides general sentiment words, while domain-specific lexicons capture context-dependent meanings. This segmentation allows the system to maintain cross-domain applicability through the general lexicon while improving accuracy through domain-specific adaptations.
Solution Approach 2:
The patent merges domain-independent sentiment lexicons with domain-specific sentiment lexicons to create a comprehensive classification system. The combination allows the system to leverage both the broad applicability of general sentiment words and the contextual accuracy of domain-specific words, resolving the contradiction between versatility and precision.
2Measurement precision
If domain-specific sentiment lexicons are created for each domain, then the sentiment classification accuracy improves, but the system complexity increases
Solution Approach 1:
The patent segments the lexicon system into modular domain-specific components that can be independently created and maintained. Each domain has its own sentiment lexicon, but they all interface with a common domain-independent framework. This segmentation reduces system complexity by allowing independent development and management of domain-specific knowledge while maintaining a unified overall structure.
Solution Approach 2:
The patent creates a universal domain-independent sentiment lexicon that serves as a foundation for multiple domain-specific lexicons. This universal component handles general sentiment words that apply across domains, reducing the need to recreate the entire lexicon for each domain and thereby reducing overall system complexity while maintaining accuracy.
3Loss of information
If domain-specific words are incorporated into the sentiment classification, then the relevance improves for that domain, but the loss of information increases when applying to other domains
Solution Approach 1:
The patent segments the sentiment lexicon into domain-specific and domain-independent portions. Domain-specific words are captured in separate lexicons that are only applied when analyzing documents from those specific domains, preventing information loss in cross-domain applications. The domain-independent lexicon maintains broad applicability without domain-specific biases.
Solution Approach 2:
The patent implements a dynamic sentiment classification system that adapts its lexicon selection based on the domain of the document being analyzed. The system dynamically switches between domain-specific and domain-independent lexicons, incorporating domain-specific words only when relevant, thereby maintaining both relevance and cross-domain applicability without permanent information loss.
Data Source
AI summary
A domain-specific sentiment classifier that can be used to score the polarity and magnitude of sentiment expressed by domain-specific documents is created. A domain-independent sentiment lexicon is established and a classifier uses the lexicon to score sentiment of domain-specific documents. Sets of high-sentiment documents having positive and negative polarities are identified. The n-grams within the high-sentiment documents are filtered to remove extremely common n-grams. The filtered n-grams are saved as a domain-specific sentiment lexicon and are used as features in a model. The model is trained using a set of training documents which may be manually or automatically labeled as to their overall sentiment to produce sentiment scores for the n-grams in the domain-specific sentiment lexicon. This lexicon is used by the domain-specific sentiment classifier.